WEBVTT 1 00:00:00.200 --> 00:00:03.680 McDonald's wraps, all flavor, no mess, no drips on your 2 00:00:03.759 --> 00:00:06.200 drip or grease on your shirts, no crumbs on your 3 00:00:06.240 --> 00:00:09.039 keyboard or sauce on your dashboard. Just all the flavor 4 00:00:09.080 --> 00:00:11.880 of sweet chili, chicken, barbecue and bacon chicken and New 5 00:00:11.960 --> 00:00:16.320 Tika Chicken wrapped up NAT served after eleven am Tika 6 00:00:16.359 --> 00:00:18.679 Chicken Wrap available until thirteenth July twenty sixth Fees applied 7 00:00:18.679 --> 00:00:20.239 to delivery orders surject to availability. 8 00:00:20.960 --> 00:00:25.320 Love MNS quality and value you can trust. At MNS 9 00:00:25.480 --> 00:00:29.039 you can update your home, your wardrobe and theirs with 10 00:00:29.239 --> 00:00:33.479 quality you can count on. There's the ultimate twelve euro lespra, 11 00:00:34.000 --> 00:00:38.520 the softest ever kid sweats from six euro fifty, luxurious 12 00:00:38.560 --> 00:00:42.560 Egyptian cotton towels from eight euro twenty five, and so 13 00:00:42.840 --> 00:00:47.759 much more value you can trust only at MNS shop 14 00:00:47.799 --> 00:00:51.439 in store or at Marsanspencer dot I e oh, hey. 15 00:00:51.359 --> 00:00:54.280 It's the bunny that you swear you saw on the lawn, 16 00:00:54.399 --> 00:00:56.640 even if no one else believes you. Ali ward and 17 00:00:56.679 --> 00:01:00.240 here's ologies Hey, am I a real person? Unfortunately? I 18 00:01:00.280 --> 00:01:03.960 am am I intelligent. That's up for debate. But this 19 00:01:04.079 --> 00:01:07.760 week we are taking a dive into artificial intelligence and 20 00:01:08.040 --> 00:01:11.760 brain data with a scholar in the matter. So listen. 21 00:01:12.159 --> 00:01:16.280 The past few months been a little surreal photoshops out there, 22 00:01:16.480 --> 00:01:19.680 generating backgrounds to cut your cousin's ex girlfriend out of 23 00:01:19.719 --> 00:01:24.519 your wedding photos, chat gpt is, writing obituaries and friendly 24 00:01:25.079 --> 00:01:29.040 a lot of horse bucky. There's also groundbreaking labor strikes 25 00:01:29.040 --> 00:01:31.640 and the arts, which we covered in the field trip 26 00:01:31.680 --> 00:01:35.000 episode from the WGA strike lines. If you haven't heard it, 27 00:01:35.079 --> 00:01:38.040 I'll link in the show notes. But I heard about 28 00:01:38.079 --> 00:01:40.719 this guest's work and I said, please, please, please talk 29 00:01:40.719 --> 00:01:44.120 to me about how to feel about AI. Are we 30 00:01:44.280 --> 00:01:48.359 farting around the portal to a new and potentially shittier 31 00:01:48.359 --> 00:01:52.560 way of living? Or will AI say hey, dipshits. I 32 00:01:52.680 --> 00:01:54.879 ran some simulations and here's what we have to do 33 00:01:54.920 --> 00:01:57.680 to unextinct you in the next century. We're going to 34 00:01:57.719 --> 00:02:01.040 find out. So this guest has studied law Dartmouth, Harvard, 35 00:02:01.239 --> 00:02:04.439 and Duke, and been a professor at Vanderbilt University and 36 00:02:04.560 --> 00:02:08.719 is now at Duke's Institute for Genome Sciences and Policy. 37 00:02:09.159 --> 00:02:12.199 She recently delivered a TED talk called Your Right to 38 00:02:12.319 --> 00:02:16.319 Mental Privacy and the Age of Brain Sensing Tech, and 39 00:02:16.719 --> 00:02:19.400 just authored a new book called The Battle for Your Brain, 40 00:02:19.599 --> 00:02:23.719 defending the right to think freely in the age of neurotechnology. 41 00:02:24.000 --> 00:02:25.960 But before we chat with her, a quick thank you 42 00:02:26.039 --> 00:02:28.439 to patrons of the show who support at patreon dot 43 00:02:28.439 --> 00:02:30.919 com slash ologies for a book or more a month 44 00:02:31.000 --> 00:02:33.639 and submit their questions for the second half. And thank 45 00:02:33.680 --> 00:02:36.879 you to everyone in ologies merch dot com, shirts and 46 00:02:36.960 --> 00:02:39.879 hats and such. Of course, you can also support the 47 00:02:39.879 --> 00:02:42.919 show just by leaving a review and I may delight 48 00:02:43.000 --> 00:02:45.280 you by reading it, such as this one left this 49 00:02:45.319 --> 00:02:49.280 week by environmental lawyer Harrison Harrison Harrison, who wrote a 50 00:02:49.360 --> 00:02:55.120 review calling Ologies and OUI guy ratituey rip Rarin good time. 51 00:02:55.400 --> 00:02:58.159 So yeah, I read them all. Thank you Harrison for that. Okay, 52 00:02:58.360 --> 00:03:02.560 neurotechnology get into this how the brain interacts with technology 53 00:03:02.599 --> 00:03:07.759 and also techno neurology, how tech is striving to replicate 54 00:03:07.879 --> 00:03:11.599 and surpass human intelligence and what that means for us all. 55 00:03:11.879 --> 00:03:16.400 So let's beat up our way into a talk about texting, scrolling, cheating, 56 00:03:16.759 --> 00:03:23.759 brain implants, mental health, doomsday scenarios, congressional hearings, apocalypse, potential 57 00:03:24.120 --> 00:03:29.840 medical advances, biometric mining, and why suddenly artificial intelligence is 58 00:03:29.919 --> 00:03:36.560 on our minds. With law professor and neurotechnologist, doctor Nita Farhani. 59 00:03:54.879 --> 00:03:56.479 Nita Fahani, she her. 60 00:03:56.560 --> 00:03:59.000 So good to meet you, terrifying to meet you. Are 61 00:03:59.080 --> 00:04:01.360 you the scariest person at a dinner party because of 62 00:04:01.439 --> 00:04:02.199 how much you know? 63 00:04:02.479 --> 00:04:05.960 I'm not. I'm not a scary person, and I find 64 00:04:05.960 --> 00:04:09.840 that people think that it's equal parts fascinating and terrifying. 65 00:04:09.840 --> 00:04:11.919 So if anything, I think I'm a great dinner gas right, 66 00:04:11.919 --> 00:04:12.879 because they're fascinated. 67 00:04:13.120 --> 00:04:16.759 I definitely should clarify that you are. There's nothing scary 68 00:04:16.759 --> 00:04:20.000 about you. The information that you hold is like, no, 69 00:04:20.160 --> 00:04:22.000 I know, do I want to look? Do I not 70 00:04:22.000 --> 00:04:23.680 want to look? Do I want to look? It's thrilling 71 00:04:24.079 --> 00:04:24.959 like a horror film. 72 00:04:25.079 --> 00:04:27.560 Yeah, it's like people can't look away. Yeah, right, And 73 00:04:27.600 --> 00:04:29.160 that's good because I don't want them to look away. 74 00:04:29.160 --> 00:04:31.000 I want them to know. But at the same time, 75 00:04:31.079 --> 00:04:33.360 what I usually get is like, wait, this is real, 76 00:04:33.600 --> 00:04:35.800 Like what you're talking about is it actually exists? And 77 00:04:35.839 --> 00:04:38.399 people are really using it and employers are really using 78 00:04:38.439 --> 00:04:41.439 it and governments are really using it and wait what Yeah? 79 00:04:41.560 --> 00:04:43.800 Do you spend a lot of your time chatting with 80 00:04:43.920 --> 00:04:47.800 people trying to warn them or calm them down? 81 00:04:47.959 --> 00:04:53.920 Yes? So On the one hand, I am trying to 82 00:04:54.519 --> 00:04:57.480 raise the alarm and to help people understand that this 83 00:04:57.600 --> 00:05:01.720 whole area of being able to decode mood and really 84 00:05:01.920 --> 00:05:05.199 hack and track the brain is a new frontier and 85 00:05:05.240 --> 00:05:08.160 the final frontier of what it means to be human 86 00:05:08.319 --> 00:05:13.360 and privacy and freedom. And at the same time, I 87 00:05:13.399 --> 00:05:17.480 don't want to make people have the reactionary approach to technology, 88 00:05:17.480 --> 00:05:20.920 which is like, Okay, then let's banmon, because the promise 89 00:05:21.079 --> 00:05:25.560 is also extraordinary, and so I am very much equal 90 00:05:25.639 --> 00:05:28.839 part like, let me help you understand not only what 91 00:05:28.839 --> 00:05:31.600 the promise is and why you're likely to adopt it, 92 00:05:32.120 --> 00:05:34.600 but why before you do so, and before we as 93 00:05:34.639 --> 00:05:37.680 a society at scale adopt this technology, that we make 94 00:05:37.720 --> 00:05:40.480 some really important choices that'll actually make it good for 95 00:05:40.600 --> 00:05:45.000 us and not the most Orwellian, frightening, scary thing possible. 96 00:05:45.040 --> 00:05:47.600 I feel like there's few topics that have this much 97 00:05:47.720 --> 00:05:52.319 true ambivalence, of so much good and so much potential 98 00:05:52.360 --> 00:05:55.879 for misuse. Did your brain become a lawyer brain because 99 00:05:55.920 --> 00:05:59.079 of those sort of like philosophical conundrums. What drew you 100 00:05:59.120 --> 00:06:01.040 to this kind of deep deep thought? 101 00:06:01.319 --> 00:06:04.360 Yeah, I've always been driven to the questions that are 102 00:06:04.399 --> 00:06:08.920 at the intersection of philosophy and science. Like in high school, 103 00:06:08.920 --> 00:06:10.720 I was really interested in the science, but I was 104 00:06:10.800 --> 00:06:14.399 a policy debater. In college, I was government minor and 105 00:06:14.720 --> 00:06:18.639 science major, and I did in lab stuff, but largely 106 00:06:19.160 --> 00:06:20.319 things that were policy. 107 00:06:20.720 --> 00:06:24.920 So Nita got several graduate degrees studying law in science, 108 00:06:25.279 --> 00:06:30.120 behavioral genetics and neuroscience, the philosophy of mind, neuroethics, bioethics, 109 00:06:30.160 --> 00:06:33.600 and even reproductive rights and policy in Kenya. And she 110 00:06:33.720 --> 00:06:37.439 said all her work seems to gravitate towards this intersection 111 00:06:37.560 --> 00:06:42.439 of philosophy and law and science because she had fundamental 112 00:06:42.519 --> 00:06:43.680 questions like do. 113 00:06:43.680 --> 00:06:46.519 We have free will? And you know, do we have 114 00:06:46.639 --> 00:06:50.279 like fundamental autonomy and freedom and how do we put 115 00:06:50.319 --> 00:06:53.120 into place the protections. But I've always been fascinated and 116 00:06:53.399 --> 00:06:56.360 really interested in the science and the technology itself. I've 117 00:06:56.399 --> 00:06:58.439 never been a luttite. I've always been somebody who's an 118 00:06:58.439 --> 00:07:01.319 early tech adopter but clearly see what the downsides are. 119 00:07:01.360 --> 00:07:03.560 At the same time, where was tech at when you 120 00:07:03.600 --> 00:07:06.879 were getting that roster of graduate degree? Where were we at? 121 00:07:06.879 --> 00:07:09.279 Were we at emails? Were we at video calls? 122 00:07:09.560 --> 00:07:12.079 Yeah? So we were not at video calls. We were 123 00:07:12.240 --> 00:07:15.279 at emails. The Internet existed, we used it. We all 124 00:07:15.319 --> 00:07:18.519 had computers, but we didn't have cell phones. I got 125 00:07:18.519 --> 00:07:21.839 my first cell phone after I graduated from college, like 126 00:07:21.879 --> 00:07:24.480 the year after, and I had a flip phone, and 127 00:07:25.120 --> 00:07:27.399 I thought that was super cool. You know, I could 128 00:07:27.399 --> 00:07:30.000 type out a text message one character at a time, 129 00:07:30.319 --> 00:07:31.079 oh T nine. 130 00:07:31.279 --> 00:07:33.639 Yeah, I had a gold medal in T nine. I 131 00:07:33.959 --> 00:07:37.040 nice could do it without even looking at the phone. 132 00:07:37.120 --> 00:07:39.319 Where I found it harder when we had a keyboard. 133 00:07:39.639 --> 00:07:41.839 Yeah, and then I had a palm pilot, like as 134 00:07:41.879 --> 00:07:44.920 the precursor to the iPhone. And then I stood in 135 00:07:44.959 --> 00:07:47.360 line the first day that the iPhone was being sold, 136 00:07:47.480 --> 00:07:49.360 and you know, got one of the first iPhones in 137 00:07:49.399 --> 00:07:52.480 my hands. So I've seen the evolution of tech, I 138 00:07:52.480 --> 00:07:54.199 guess as I was getting all of those degrees. 139 00:07:54.439 --> 00:07:58.079 And what about in terms of neurotechnology. Have you seen 140 00:07:58.240 --> 00:08:02.519 kind of an exponential growth path in terms of technology? 141 00:08:02.839 --> 00:08:06.519 Is that growth pattern still valid or have we surpassed it? 142 00:08:06.759 --> 00:08:12.040 Slowly? Over the past decade or two, neurotechnology has been 143 00:08:12.079 --> 00:08:14.079 getting better, and the ways in which neurotech has been 144 00:08:14.120 --> 00:08:17.480 getting better has largely been kind of hardware based, which 145 00:08:17.560 --> 00:08:21.040 is the centers are getting better. Sometimes the software has 146 00:08:21.079 --> 00:08:23.399 been getting better, to be able to filter out noise, 147 00:08:23.920 --> 00:08:27.079 the algorithms, to be able to pick up brain activity 148 00:08:27.600 --> 00:08:32.879 without having muscle twitches or eyeblinks or interference from the environment, 149 00:08:32.960 --> 00:08:35.399 to pick up different information. All that's been getting better. 150 00:08:35.480 --> 00:08:40.159 But suddenly we've gone from what was improvements to just 151 00:08:40.240 --> 00:08:44.360 the past five years seeing much more rapid advances. Generative 152 00:08:44.399 --> 00:08:48.080 AI is making things move in these seismic shifts, like 153 00:08:48.120 --> 00:08:51.639 where you suddenly have just a massive leap and capabilities. 154 00:08:51.879 --> 00:08:54.960 Just real quick, before we descend into the abyss of 155 00:08:55.000 --> 00:08:59.960 ethics and possible scenarios, what is generative AI? What is AA? 156 00:09:00.399 --> 00:09:03.440 And what's just a computer computing? Okay, I look this 157 00:09:03.519 --> 00:09:05.639 up for us, and then I took a nap because 158 00:09:05.639 --> 00:09:07.840 it was confusing, and then I tried again, and here's 159 00:09:07.879 --> 00:09:10.919 what I suessed out. So artificial means it's coming from 160 00:09:10.960 --> 00:09:14.480 a machine or software and intelligence fuck, I mean that 161 00:09:14.519 --> 00:09:17.320 depends on who you ask, but broadly it means a 162 00:09:17.360 --> 00:09:23.759 capacity for logic, understanding, learning, reasoning, problem solving, and retaining facts. 163 00:09:24.159 --> 00:09:27.720 So some examples of AI are googling or search engines, 164 00:09:28.120 --> 00:09:31.600 the software that recommends other things you might like to purchase, 165 00:09:31.919 --> 00:09:36.919 navigating via self driving cars. Your Alexa understanding when you 166 00:09:36.960 --> 00:09:40.840 scream Alexa stop because she tried to get YouTube subscribe 167 00:09:40.840 --> 00:09:44.799 to Amazon Prime Music again. It also includes computers being 168 00:09:44.879 --> 00:09:49.559 chess nerds that's AI and generating artwork. And according to 169 00:09:49.600 --> 00:09:53.399 some experts, AI can be separated into a few categories, 170 00:09:53.440 --> 00:09:57.320 including on the base level, reactive machines and those use 171 00:09:57.519 --> 00:10:00.799 existing information but they don't store or learning. Then there's 172 00:10:00.879 --> 00:10:04.600 limited memory AI that can use precedent to learn what 173 00:10:04.759 --> 00:10:08.120 choices to make. There's something called theory of mind AI 174 00:10:08.440 --> 00:10:11.120 and that can try to figure out the intentions of 175 00:10:11.159 --> 00:10:14.639 a user or even acknowledged their feelings, like if you've 176 00:10:14.639 --> 00:10:17.600 ever told Alexa to get bent in a lot of 177 00:10:17.639 --> 00:10:20.120 other words and then she sassas you back. If you'd 178 00:10:20.159 --> 00:10:22.639 like to tell me how I can improve, try saying 179 00:10:22.919 --> 00:10:25.919 I have feedback. There's also a type called self aware 180 00:10:26.000 --> 00:10:30.120 AI that reflects on its own actions, and then fully 181 00:10:30.159 --> 00:10:33.240 autonomous is kind of the deluxe model of AI, and 182 00:10:33.279 --> 00:10:36.320 that just that does its own thing, that sets its 183 00:10:36.360 --> 00:10:38.679 own goals, set it and forget it if you can. 184 00:10:39.000 --> 00:10:41.480 So when did things start speeding up? When did they 185 00:10:41.519 --> 00:10:44.879 start careening toward the future like this, When computers got 186 00:10:44.919 --> 00:10:48.840 faster and smaller and better in the last ten but 187 00:10:48.960 --> 00:10:51.960 really kind of two or three years. So better hardware 188 00:10:52.200 --> 00:10:56.039 means more processing power. There's also cloud storage, and that 189 00:10:56.039 --> 00:10:59.080 adds up to something called deep learning, which kind of 190 00:10:59.080 --> 00:11:04.759 sounds creepy like a hypervigilant mannequin, but deep refers to 191 00:11:05.480 --> 00:11:08.519 many layers of networks that use what look like these 192 00:11:08.559 --> 00:11:11.919 complicated flow charts to decide what actions to take based 193 00:11:11.960 --> 00:11:14.159 on previous learning. So that's kind of what led up 194 00:11:14.159 --> 00:11:19.440 to these startlingly human like generative AI outputs and deep 195 00:11:19.480 --> 00:11:22.519 fakes where they can just straight up put Keanu Reef's 196 00:11:22.519 --> 00:11:25.759 face on your mom and then confuse the the Jesus 197 00:11:25.759 --> 00:11:29.039 out of me on TikTok or chat GBT, which is 198 00:11:29.120 --> 00:11:34.919 one language model. Chatbot computers are starting to pass bar exams. 199 00:11:35.320 --> 00:11:38.960 Maybe they're writing the Quippi flirtations on your dating app. 200 00:11:39.080 --> 00:11:42.519 Who knows. Meanwhile, less than one hundred years ago, a 201 00:11:42.559 --> 00:11:45.440 lot of the US didn't have flush toilets. In case 202 00:11:45.639 --> 00:11:48.320 you feel weird about how weird this feels, because it 203 00:11:48.440 --> 00:11:52.759 is weird. Evolutionarily, our flabby, beautiful little brains can barely 204 00:11:52.799 --> 00:11:55.519 handle the shock of a clean river coming out of 205 00:11:55.519 --> 00:11:59.320 a garden. Hose let alone some metal and rocks that 206 00:11:59.399 --> 00:12:03.480 are computers that we're training to potentially kill us. We 207 00:12:03.519 --> 00:12:04.399 don't know how to deal with that. 208 00:12:04.799 --> 00:12:09.360 So pattern recognition using machine learning algorithms has really pushed 209 00:12:09.399 --> 00:12:12.000 things forward rapidly, like a lot of brain data that 210 00:12:12.039 --> 00:12:16.039 happens in characteristic patterns and those associations between like what 211 00:12:16.080 --> 00:12:19.039 does a person seeing or hearing or thinking? How are 212 00:12:19.039 --> 00:12:21.679 they feeling? Are they tired, are they happy? Are they sad? 213 00:12:21.720 --> 00:12:24.720 Or are they stressed? Those things have been correlated with 214 00:12:24.919 --> 00:12:28.360 huge data sets and processed using machine learning algorithms in 215 00:12:28.399 --> 00:12:29.279 ways that weren't. 216 00:12:29.000 --> 00:12:30.919 Possible before I can read your mind. 217 00:12:31.240 --> 00:12:34.799 Then you have generative AI and chat GBT that enters 218 00:12:34.840 --> 00:12:39.600 the scene in November, and all of a sudden, the 219 00:12:39.639 --> 00:12:42.799 papers that are coming out are jaw dropping. Data that's 220 00:12:42.840 --> 00:12:46.000 being processed by generative AI to reconstruct what a person 221 00:12:46.080 --> 00:12:51.639 is thinking or hearing or imagining or seeing. Is next level, right, 222 00:12:51.679 --> 00:12:53.840 I mean, you know, my book came out March fourteenth, 223 00:12:54.039 --> 00:12:57.120 twenty twenty three. All of a sudden, what was happening 224 00:12:57.320 --> 00:13:02.639 was continuous language decoding from the in like really really 225 00:13:02.720 --> 00:13:05.799 high resolution using GPT one, not even the most advanced 226 00:13:05.799 --> 00:13:09.879 GPT four visual reconstruction of images that a person is 227 00:13:09.879 --> 00:13:13.120 seeing in ways that were much more precise than anything 228 00:13:13.120 --> 00:13:17.679 that we had seen previously. And that's happening at this clip. 229 00:13:17.720 --> 00:13:21.240 That is just, I think extraordinary. It's just so much 230 00:13:21.279 --> 00:13:24.159 faster than I even I would have imagined, and even 231 00:13:24.200 --> 00:13:26.600 I could have anticipated, even having written a book about 232 00:13:26.600 --> 00:13:27.600 the topic, that was. 233 00:13:27.759 --> 00:13:30.399 Literally going to be my next question, because when a 234 00:13:30.440 --> 00:13:33.200 person writes a book that doesn't happen every night. Even 235 00:13:33.240 --> 00:13:35.759 working on this book, probably for a couple of years, 236 00:13:36.200 --> 00:13:39.519 did you have any idea that your book would be 237 00:13:39.600 --> 00:13:43.720 so closely timed to such a giant leap in terms 238 00:13:43.759 --> 00:13:47.919 of public perception and awareness of AI. I mean, couldn't 239 00:13:47.919 --> 00:13:49.279 have timed it better? Well? 240 00:13:49.320 --> 00:13:52.279 I mean, of course, I'm a futurist. I was predicting 241 00:13:52.320 --> 00:13:55.639 it perfectly right now, No, I mean I wish right 242 00:13:55.759 --> 00:13:57.639 In truth, my book is like a year and a 243 00:13:57.679 --> 00:13:59.679 half late from when I was supposed to turn it 244 00:13:59.720 --> 00:14:02.200 into the editor and to the publisher. But you know, 245 00:14:02.240 --> 00:14:04.279 there was a global pandemic that got in the way 246 00:14:04.360 --> 00:14:06.440 and a bunch of other things. But I'm grateful that 247 00:14:06.759 --> 00:14:10.120 it didn't happen sooner, because I was both able to 248 00:14:10.159 --> 00:14:12.879 be part of what is a growing conversation about the 249 00:14:12.879 --> 00:14:15.639 capabilities of AI, and to see when you say to 250 00:14:15.679 --> 00:14:18.679 a person like, oh yeah, also AI can decode your brain. 251 00:14:19.559 --> 00:14:21.320 You know, that really puts a fine point on it 252 00:14:21.360 --> 00:14:23.960 for people to understand how quickly these advances are coming 253 00:14:24.519 --> 00:14:27.960 and to see how it's changing everything in society, not 254 00:14:28.000 --> 00:14:30.600 just how people are writing essays or writing emails, but 255 00:14:30.639 --> 00:14:35.480 fundamentally unlocking the mysteries of the mind that people never 256 00:14:35.519 --> 00:14:38.840 thought before possible, and the risk that that opens up, 257 00:14:38.879 --> 00:14:42.240 and the possibilities of mental manipulation and hacking and tracking. 258 00:14:42.879 --> 00:14:46.879 Those are dangers that I think a year ago, before 259 00:14:46.919 --> 00:14:50.399 people really woke up to the risks of AI, they 260 00:14:50.440 --> 00:14:52.559 would not have been having the conversation in the same 261 00:14:52.559 --> 00:14:54.240 way that they are around the book. And now they 262 00:14:54.279 --> 00:14:57.440 are having that conversation seeing the broader context and seeing 263 00:14:57.919 --> 00:15:01.360 the alarm bells everywhere. Right, oh wait, we really do 264 00:15:01.440 --> 00:15:03.840 need to regulate or recognize them rights or do something. 265 00:15:04.159 --> 00:15:09.000 So futurists are urging some foresight. Congressional panels have aired 266 00:15:09.039 --> 00:15:11.360 on c SPAN and there seems to be this kind 267 00:15:11.399 --> 00:15:15.840 of collective side eye and like a hope someone's on 268 00:15:15.879 --> 00:15:16.879 top of this, right. 269 00:15:16.960 --> 00:15:19.039 So I mean, I think people are looking for some 270 00:15:19.120 --> 00:15:22.559 guidance and to have somebody come at it from a 271 00:15:22.600 --> 00:15:24.639 balanced perspective, like, wait a minute, there's a lot of 272 00:15:24.679 --> 00:15:27.639 good here, and there's some serious risks, and here's a 273 00:15:27.679 --> 00:15:31.279 potential pathway forward. I think, you know, instead of like pause, 274 00:15:31.360 --> 00:15:33.600 which everybody says like of course, we can't just pause, 275 00:15:33.840 --> 00:15:37.679 or you know, a doomsday scenario without any positive like oh, 276 00:15:37.759 --> 00:15:40.399 let's regulate AI, I think we need voices at the 277 00:15:40.440 --> 00:15:42.639 table who are thinking about it both in a balanced way, 278 00:15:42.679 --> 00:15:45.360 but also are coming forward with like here's some concrete 279 00:15:45.399 --> 00:15:48.000 things we could do right now that would actually help 280 00:15:48.039 --> 00:15:48.559 the problem. 281 00:15:48.799 --> 00:15:52.320 So we know a few types of AI, from googling 282 00:15:52.600 --> 00:15:57.279 a source for a research paper or digitally removing your 283 00:15:57.320 --> 00:16:01.679 cousins X from your wedding photos. But about technology that's 284 00:16:01.759 --> 00:16:04.480 gathering data from our brains. 285 00:16:04.480 --> 00:16:09.759 Let me give you the spectrum. There's medical grade neurotechnology. 286 00:16:09.799 --> 00:16:14.279 This is technology that people might imagine in a doctor's 287 00:16:14.279 --> 00:16:18.200 office where somebody puts on an EEG electro encepholography cap 288 00:16:18.720 --> 00:16:21.440 that has a bunch of different wires coming out of 289 00:16:21.480 --> 00:16:23.720 it and a bunch of gel that's applied to their head, 290 00:16:23.759 --> 00:16:26.279 and a bunch of sensors. That's picking up electrical activity, 291 00:16:26.320 --> 00:16:28.159 which we'll get back to in a minute. Then there's 292 00:16:28.200 --> 00:16:32.240 the clunky giant machine, a functional magnetic resonance imaging machine 293 00:16:32.279 --> 00:16:37.080 which can peer deeply into the brain. And somebody might 294 00:16:37.120 --> 00:16:39.960 have already undergone an fMRI test for something like a 295 00:16:39.960 --> 00:16:42.639 brain tumor to kind of look more deeply into the brain. 296 00:16:42.679 --> 00:16:45.759 And what that's picking up is changes in blood flow 297 00:16:45.799 --> 00:16:49.440 across the brain, which tells us something about different areas 298 00:16:49.480 --> 00:16:52.559 that are activated anyto particular time and what those patterns 299 00:16:52.600 --> 00:16:53.000 might mean. 300 00:16:53.320 --> 00:16:56.279 So if you've never had an MRI, I guess, congratulations, 301 00:16:56.320 --> 00:16:59.879 that's probably good. But this is magnetic resonance imaging and 302 00:17:00.320 --> 00:17:04.119 it's pretty exciting how these strong as magnets all line 303 00:17:04.200 --> 00:17:06.759 up the hydrogen atoms in your body to go one 304 00:17:06.799 --> 00:17:09.799 direction and then they release them and from that they 305 00:17:09.799 --> 00:17:13.720 can see inside of your boudet. Now, an fMRI is 306 00:17:13.880 --> 00:17:17.039 a functional MRI, and to put it in super simple terms, 307 00:17:17.119 --> 00:17:20.359 it's kind of like animation instead of a still picture, 308 00:17:20.440 --> 00:17:23.359 but it's of your brain. So when you see imaging 309 00:17:23.440 --> 00:17:28.400 examples of how someone's melon illuminates like a Christmas tree 310 00:17:28.440 --> 00:17:32.559 to a certain stimuli, that's fMRI technology tracking blood flow to 311 00:17:32.680 --> 00:17:35.759 different regions of the brain. And this fMRI technology is 312 00:17:35.839 --> 00:17:38.960 used in a lot of neuro and psychology research. 313 00:17:39.240 --> 00:17:42.359 And then there's something like functional near infrared spectroscopy, which 314 00:17:42.400 --> 00:17:46.359 is more portable and it's also measuring changes in the brain, 315 00:17:46.400 --> 00:17:49.839 but it's using optical and infrared lights in order to 316 00:17:49.880 --> 00:17:50.240 do so. 317 00:17:50.599 --> 00:17:55.640 And that functional near infrared spectroscopy looks for changes in 318 00:17:55.880 --> 00:18:00.799 oxyhemoglobin and deoxyhemoglobin in the brain. It might not matter 319 00:18:00.839 --> 00:18:03.599 to you right now as you're cleaning your shower grout 320 00:18:03.720 --> 00:18:06.559 or your carpooling, but in clinical settings it comes in 321 00:18:06.599 --> 00:18:11.000 handy for patients with strokes, or learning about Alzheimer's or Parkinson's, 322 00:18:11.079 --> 00:18:14.200 or even anxiety or a traumatic brain injury, which my 323 00:18:14.200 --> 00:18:16.960 brain would like you to know I've had, and I 324 00:18:17.000 --> 00:18:20.720 will link the traumatic brain injury or the neuropathology episode 325 00:18:20.720 --> 00:18:23.440 about my hell and Narnar concussion I got last year 326 00:18:23.440 --> 00:18:25.720 in the show notes. But yes, there are a lot 327 00:18:25.759 --> 00:18:28.720 of ways to get data from a brain, including T 328 00:18:28.880 --> 00:18:32.880 scans and PET scans with radioactive tracers. But what about 329 00:18:33.039 --> 00:18:36.000 non medical uses? Do they exist? Oh? 330 00:18:36.000 --> 00:18:38.240 Boy, how do you do they? If you then look 331 00:18:38.240 --> 00:18:41.039 at what's happening in the consumer space. And the consumer space, 332 00:18:41.319 --> 00:18:44.559 you take the sixty four one hundred and twenty electrodes 333 00:18:44.559 --> 00:18:45.920 that are in a big cap, and then you have 334 00:18:45.960 --> 00:18:48.960 a couple of them that are put into a forehead 335 00:18:49.000 --> 00:18:52.720 band or a baseball cap, or increasingly, what's coming is 336 00:18:52.799 --> 00:18:55.799 brain sensors that are embedded in everyday technology. So you 337 00:18:55.839 --> 00:18:59.359 and I are both wearing headphones, and the soft cups 338 00:18:59.400 --> 00:19:02.000 that go around our ears are being packed with sensors 339 00:19:02.039 --> 00:19:05.319 that can pick up brain activity by reading the electrical 340 00:19:05.359 --> 00:19:08.599 activity through our scalp. You want my tinfoil hat, or 341 00:19:08.640 --> 00:19:11.480 if we were wearing earbuds inside of our ears, instead 342 00:19:12.000 --> 00:19:14.720 embedding brain sensors inside of those that can pick up 343 00:19:14.720 --> 00:19:18.079 electrical activity in our brain activity as we're thinking or 344 00:19:18.119 --> 00:19:22.119 doing anything, and those become much more familiar and much 345 00:19:22.200 --> 00:19:24.720 more commonplace very quickly. So there's just a few of 346 00:19:24.720 --> 00:19:27.240 those products that are on the market, but that's where 347 00:19:27.279 --> 00:19:29.279 most of the big tech companies are going, is to 348 00:19:29.319 --> 00:19:32.960 embed brain sensors into everyday devices like earbuds and headphones 349 00:19:32.960 --> 00:19:36.200 and even watches that pick up brain activity from your 350 00:19:36.240 --> 00:19:38.680 brain down your arm to your wrist and picks up 351 00:19:38.680 --> 00:19:40.960 your intention to move or to type or to swipe 352 00:19:41.000 --> 00:19:41.880 or something like that. 353 00:19:42.000 --> 00:19:45.839 So to use like a medical analogy. You know, continuous 354 00:19:45.960 --> 00:19:49.839 glucose monitors. These are a powerful tool for diabetics to 355 00:19:49.880 --> 00:19:52.319 monitor their blood sugar levels and their insulin needs, and 356 00:19:52.359 --> 00:19:54.559 we cover those in the two part Dibetology episode with 357 00:19:54.559 --> 00:19:58.279 doctor Mike Natter. But now continuous glucose monitors are starting 358 00:19:58.279 --> 00:20:01.160 to become available to people without diabetes just to better 359 00:20:01.279 --> 00:20:06.279 understand their metabolisms and their dietary responses, their mood and energy. 360 00:20:06.519 --> 00:20:10.119 So all of these neuroimaging and all this data was 361 00:20:10.400 --> 00:20:13.480 just used in clinical and research settings by people in 362 00:20:13.720 --> 00:20:18.000 crisp coats carrying metal clipboards, but it's starting to pop 363 00:20:18.079 --> 00:20:20.079 up in the market. Now. This is great news, right 364 00:20:20.480 --> 00:20:24.000 understanding your brain? Yeah yeah, But not all the research 365 00:20:24.119 --> 00:20:28.039 in consumer applications is solid, and some make some wild 366 00:20:28.160 --> 00:20:32.759 claims of efficacy. Others argue that if a device can 367 00:20:33.000 --> 00:20:37.920 enhance our moods and sharpen us cognitively but cost some 368 00:20:38.319 --> 00:20:42.440 serious cash, doesn't that just widen a privileged gap even further? 369 00:20:42.880 --> 00:20:46.039 But I guess so does college. I don't know. In 370 00:20:46.119 --> 00:20:48.839 the US, you need a GoFundMe to pay for chemo, 371 00:20:49.079 --> 00:20:52.640 so we've got a lot of pretty large systemic fish 372 00:20:52.680 --> 00:20:56.160 to fry. But if you've got money, you can buy 373 00:20:56.480 --> 00:21:00.279 EEG headsets that track your mood and emotions and stress 374 00:21:00.279 --> 00:21:03.119 for a few grand. There's others that track your heart 375 00:21:03.160 --> 00:21:06.920 rate and brain waves for sleep and meditation. There are 376 00:21:07.079 --> 00:21:10.519 VR gaming sets that track brain waves, and even a 377 00:21:10.559 --> 00:21:13.519 matel game called mind Flex you can buy for like 378 00:21:13.559 --> 00:21:14.799 one hundred and twenty bucks. 379 00:21:15.119 --> 00:21:19.200 But Nita says, all of those consumer based technologies pick 380 00:21:19.279 --> 00:21:21.720 up a little bit of like kind of low resolution 381 00:21:21.839 --> 00:21:25.160 information right now. They pick up if you're stressed, if 382 00:21:25.200 --> 00:21:28.240 you're happy, or if you're sad, if you're tired, like 383 00:21:28.279 --> 00:21:30.480 it maybe picks up that your mind is wandering and 384 00:21:30.519 --> 00:21:33.440 you're you know, kind of like dozing off. And the 385 00:21:33.880 --> 00:21:39.119 things like fMRI pick up much more precise information. Now 386 00:21:39.119 --> 00:21:41.480 that could just be a matter of time. It could 387 00:21:41.480 --> 00:21:44.359 be that as machine learning algorithms and gender tove AI 388 00:21:44.440 --> 00:21:46.519 gets up and get applied to the electrical activity in 389 00:21:46.559 --> 00:21:49.119 the brain, that it'll get better and better and better. 390 00:21:49.319 --> 00:21:51.880 And it's interesting because in a way you could think 391 00:21:51.880 --> 00:21:56.359 about AI as being the convergence between computer science and neuroscience. 392 00:21:56.559 --> 00:22:02.880 So computer scientists have been designing algorithms that can process 393 00:22:02.960 --> 00:22:05.640 information in very narrow ways and they're very good at 394 00:22:05.640 --> 00:22:09.319 doing specific tasks. So, for example, a doctor or a 395 00:22:09.319 --> 00:22:14.200 pathologist who's looking at many different samples of tissue to 396 00:22:14.240 --> 00:22:17.200 figure out if it looks cancerous or not can only 397 00:22:17.240 --> 00:22:20.559 see so many samples in a lifetime, and so they've 398 00:22:20.599 --> 00:22:23.640 marked them and labeled the data, and a machine learning 399 00:22:23.680 --> 00:22:27.000 algorithm can be trained on that data, which is like, 400 00:22:27.359 --> 00:22:31.119 here's thousands of images that are cancer and not cancer. 401 00:22:31.920 --> 00:22:35.079